The challenge
A Structural Problem Hiding Inside a Process
Every time a bank credit arrived from a parent entity, the company's finance team faced the same situation. The payment came with no invoice reference. No remittance advice. No indication of which tower, which child entity, or which invoice it was meant to settle.
So the team did what it had always done. It split the credit from memory, drawing on years of institutional knowledge to allocate amounts across entities and invoices. Most of the time, that worked. When it didn't, the credit sat on-account and waited for the end of the quarter. At scale, this meant late weekends at every month end, repeated back-and-forth with customers to chase remittance details, and hours spent manually resolving and posting entries that should have cleared the day the money arrived.
The problem was not the occasional misallocation. The problem was structural. A large share of the group's customers are multi-entity businesses, and their payments rarely map one-to-one to an invoice. A parent entity pays for invoices raised on its subsidiaries. One child settles what another child owes. A single receipt has to be split across several entities and several invoices at once. Without an invoice-level view, the team could not tell which invoice a receipt actually closed, and closing out a specific invoice became a persistent challenge. Reconciliation happened once a quarter, so the ledger ran a quarter behind reality. DSO overstated the position. Every operational decision made against the AR book was made against numbers that were, by definition, out of date.
The foundation
The ERP Remains the Book of Records
The ERP was not the problem. It did exactly what it was built to do: it recorded what happened. It knew that the cash landed in the bank. It did not know who sent it, which customer entity the payment belonged to, or which invoices across that customer's hierarchy it was meant to settle. That knowledge lived in people. It could not be automated. It could not scale.
The team was not failing. It was carrying a structural limitation that no amount of process discipline could fully resolve. The ERP is the book of records. What was missing was a reasoning layer that could work above it.
The solution
What Alfred Did
The team wrote its strategy for all cash allocation in plain English. Alfred combined that strategy with its own AR knowledge and the payment details, applied the allocation, and delivered the invoice-level view and posting to the team. Alfred AR works above the ERP, which remains the book of records.
When a bank credit arrived with no remittance advice, Alfred identified the payer at 95% confidence or better, treated the parent code as covering the full parent-plus-children structure, and surfaced every open child-entity invoice for exact allocation. The guesswork was gone. What remained for the team was one decision: approve or correct.
Speed followed from identification. What previously sat on-account for weeks cleared the day the money landed. Alfred posted cash at the invoice level, bill-to-bill, same day.
When no remittance existed, Alfred did not wait for one. It emailed the identified customer within seconds of the credit landing, included the payment reference and all open invoices, and stored the customer's reply as the official remittance advice. A follow-up cycle that once ran for days was reduced to minutes.
Every action wrote back to the ERP as the system of record. Nothing was migrated. The company's existing infrastructure remained intact.
One payment. The right invoices.
- 01Payment arrives
A bank credit, without an invoice reference.
- 02Alfred allocates
Your strategy, applied across the customer’s entities.
- 03Your team reviews
One decision: approve or correct.
- 04ERP updated
Cash posted against the invoices it closes.
Your ERP remains the book of records.
The results
What Changes When the Lag Is Gone?
Today, 80% of cash is posted the same day it arrives, at the invoice level. The team can see exactly which invoice each payment closed. The collections team, which had spent a significant share of its time pursuing accounts that had already settled, now works a queue of real risk. The group AR figure is accurate on the day it is reported.
The structural complexity did not disappear. Bank credits still arrive with no remittance advice. Parent entities still pay across child structures. What changed is how the team handles that complexity: 80% of cash now posts the same day, with a clear view of the invoices each payment closes.
The company did not change its ERP. It did not hire additional headcount. The resolution came from adding a reasoning layer above what was already there.
The credit still lands with no reference. Alfred still knows where it goes.

